Answers are grounded in this paper and cite their sources.
We study how bidders adjusted to a switch from second-price to first-price auction rules on a large online advertising exchange. Under first-price rules, bidders should strategically reduce (“shade”) their bids. To assess whether observed bid reductions were sufficient given the competition each bidder faced, we develop a nonparametric estimator of a lower bound on the valuation underlying each bid. The estimator relies on localized measures of competition and does not require assumptions of bidder symmetry, valuation independence, or equilibrium bidding. Applying the method to more than 40 million auctions involving eight major demand-side platforms, we find pervasive insufficient bid shading after the switch. The implied lower bounds on valuations remain substantially above pre-switch levels for at least three months after the transition, indicating that none of the bidders we study shaded sufficiently. This finding is robust to alternative definitions of local competition, richer conditioning information, and variations in the assumptions underlying the estimator. The results suggest that adjustment to major auction-design changes can be surprisingly slow even among sophisticated algorithmic bidders, with implications for marketplace design and the interpretation of short-run “A/B test” experiments.
No takes yet. Share an insight, caveat, or question.
Zeithammer et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: